Update app.py

app.py

@@ -9,11 +9,11 @@ import json
 
 # Load the Vicuna 7B model and tokenizer
 vicuna_tokenizer = AutoTokenizer.from_pretrained("lmsys/vicuna-7b-v1.3")
-vicuna_model = AutoModelForCausalLM.from_pretrained("lmsys/vicuna-7b-v1.3")
+vicuna_model = AutoModelForCausalLM.from_pretrained("lmsys/vicuna-7b-v1.3", load_in_4bit=True)
 
 # Load the LLaMA 7b model and tokenizer
 llama_tokenizer = AutoTokenizer.from_pretrained("daryl149/llama-2-7b-chat-hf")
-llama_model = AutoModelForCausalLM.from_pretrained("daryl149/llama-2-7b-chat-hf")
+llama_model = AutoModelForCausalLM.from_pretrained("daryl149/llama-2-7b-chat-hf", load_in_4bit=True)
 
 os.environ['OPENAI_API_KEY']
 openai.api_key = os.environ['OPENAI_API_KEY']